Field of the Invention
This invention relates to an antibody variant composition in which, among N-glycoside-linked sugar chains which are bound to the Fc region of an antibody, sugar chains which are bound to Asn residues at position other than position 297 according to the EU index are decreased or deleted; an antibody variant composition in which, among N-glycoside-linked sugar chains which are bound to the Fc region of an antibody, extra sugar chains which are bound to Asn residues at positions other than position 297 according to the EU index are decreased or deleted and effector activity of the antibody is maintained; a DNA encoding the antibody variant molecules; a cell capable of producing the antibody variant composition, a process for producing the antibody variant composition; and a method for reducing extra sugar chains which are bound to Asn residues at position other than position 297 according to the EU index among the N-glycoside-linked sugar chains which bind to the Fc region of the antibody.
Brief Description of the Background Art
Since antibodies have high binding activity and binding specificity and high stability in blood, their applications to the diagnostic, preventive and therapeutic agents of various human diseases have been attempted (Non-patent Document 1). In addition, human chimeric antibodies or humanized antibodies were prepared from non-human animal antibodies making use of recombinant techniques (Non-patent Documents 2 to 5).
A human chimeric antibody comprises an antibody variable region of a non-human animal antibody and a constant region of a human antibody. A humanized antibody is an antibody in which complementarity determining regions (hereinafter referred to as CDR) of a human animal antibody is substituted with CDRs of a non-human antibody.
The human chimeric antibodies and humanized antibodies resolved problems possessed by non-human animal antibodies such as mouse antibodies, such as the high immunogenicity, low effector function and short blood half-life, and applications of monoclonal antibodies to pharmaceutical preparations were made possible by using them (Non-patent Documents 6 to 9). In the Unites States, for example, a plural of humanized antibodies have already been approved as an antibody for cancer treatment, and are on the market (Non-patent Document 10).
These human chimeric antibodies and humanized antibodies actually show effects to a certain degree at clinical level, but therapeutic antibodies having higher effects are required.
For example, in the case of single administration of Rituxan (registered tradename) (Non-patent Document 11) (manufactured by MEC/Roche/Genentech) which is a human chimeric antibody to CD20, it was reported that its response rate for recurrent low malignancy non-Hodgkin lymphoma patients in the phase III clinical test is no more than 48% (complete remission 6%, partial remission 42%), and its average duration of response is 12 months (Non-patent Document 12).
In the case of combination use of Rituxan (registered tradename) and chemotherapy (CHOP: Cyclophosphamide, Doxorubicin, Vincristine), it was reported that its response ratio for recurrent low malignancy and follicular non-Hodgkin lymphoma patients by the phase II clinical test is 95% (complete remission 55%, partial remission 45%), but side effects due to CHOP were found (Non-patent Document 13).
In the case of single administration of Herceptin (registered tradename) (manufactured by Genentech) which is a humanized antibody to HER2, it was reported that its response ratio for metastatic breast cancer patients in the phase III clinical test is only 15%, and its average duration of response is 9.1 months (Non-patent Document 14).
The human antibody molecule is also called immunoglobulin (hereinafter referred to as Ig) and classified into isotypes of IgA, IgD, IgE, IgG, and IgM based on its molecular structure.
An antibody molecule of human IgG (hereinafter referred to as IgG) type which is mainly used as a therapeutic antibody is an antibody molecule comprises two heavy chains (hereinafter referred to as H chain) and two light chains (hereinafter referred to as L chain).
An H chain comprises individual domain structures which are respectively called an H chain variable region (hereinafter referred to as VH), CH1, hinge, CH2 and CH3 domains from the N-terminal. Also, the CH2 domain and CH3 domain in combination are called Fc region.
An L chain comprises individual domain structures which are respectively called an L chain variable region (hereinafter referred to as VL) and an L chain constant region (hereinafter referred to as CL) from the N-terminal.
The H chain of the IgG type antibody includes four subclasses consisting of IgG1, IgG2, IgG3 and IgG4. The H chains of the individual IgG subclasses are homologous with each other at almost 95% amino acid sequence identity in the constant region, except for highly-variable hinges.
Although individual IgG subclasses have highly homologous with each other in the amino acid sequence, they have different biological activity in strength (Non-patent Document 15). Examples of the biological activities include an effector function such as a complement-dependent cytotoxicity (hereinafter referred to as CDC activity), an antibody-dependent cellular cytotoxicity (hereinafter referred to as ADCC activity) and a phagocytic activity. These activities play an important role in eliminating foreign substances and pathogens in the living body.
A family of Fcγ receptor (hereinafter referred to as FcγR) is expressed on the surface of various leukocytes such as natural killer cells (hereinafter referred to as NK cell), monocytes, macrophages, granulocytes and the like.
FcγR is classified into active type FcγR such as FcγRI, FcγRIIa, FcγRIIIa and FcγRIIIb and inhibitory type FcγR such as FcγRIIb. IgG antibodies, particularly IgG1 and IgG3 in human, strongly bind to these receptors so that they induce ADCC activity and phagocytosis by leucocytes.
ADCC activity means a cell lysis reaction which is generated by cytotoxic molecules released from NK cell, such as perforin and granzyme, as a result that an antibody bound to an antigen mainly binds to FcγRIIIa on the NK cell surface via the Fc moiety (Non-patent Documents 16 and 17). In general, ADCC activity shows an order of IgG1>IgG3>>IgG4 IgG2 (Non-patent Documents 18 and 19). A binding site to FcγRIIIa on the Fc of the antibody is present in CH2 domain, and a crystal structure analysis shows that one molecule of FcγRIIIa binds to the space between two CH2 domains (Non-patent Document 20).
CDC activity means a reaction in which an antibody bound to its antigen activates reaction cascade of a group of serum proteins called complement system and finally lyses the target cell. CDC activity is high in human IgG1 and IgG3 and generally shows an order of IgG3≥IgG1>>IgG2≈IgG4. The complement system is classified into respective components of C1 to C9, and most of them are enzyme precursors which express the enzyme activity by partial degradation.
CDC activity induces the cell lysis reaction by starting from the binding of C1q as one component of C1 to the Fc region of an antibody on the target cell, advancing activation cascade by partial degradation of respective component by the components of the previous step, and finally forming a pore forming unit which is called membrane attack complex formed by C5 to C9 on the cell membrane of target cell (Non-patent Documents 16 and 17). An Fc region amino acid substitution study suggests that the binding site of C1q on the Fc region is present in the CH2 domain (Non-patent Document 21)
Importance of the above-mentioned effector function is recognized also in the drug effect mechanism of a therapeutic antibody to be used in clinical practice. The above-mentioned Rituxan (registered trademark) is a human chimeric antibody of IgG1 subclass and exhibits ADCC activity and CDC activity in vitro (Non-patent Document 22).
Further, regarding its clinical effect, Rituxan (registered trademark) is suggested to actually exhibit the effector function in the human body of patients, since Rituxan (registered trademark) exhibits high therapeutic effect in patients having FcγRIIIa genotype which shows strong ADCC activity (Non-patent Document 23), the complement components in blood are quickly consumed after administration (Non-patent Document 24), expression of CD59 which is a factor to inhibit CDC activity is increased in the cancer cells of patients who recurred after administration (Non-patent Document 25), and the like.
Herceptin (registered trademark) is also an IgG1 subclass humanized antibody, and it was reported to exhibit high therapeutic effects in patients having FcγRIIIa genotype which shows strong ADCC activity (Non-patent Document 26).
In addition, since human IgG escaped from degradation by lysosome by binding to a neonatal Fc receptor (neonatal Fc receptor for IgG; FcRn, hereinafter referred to as FcRn) which is expressed on such as vascular endothelial cell, under low pH conditions in endosome, it has a long blood half-life for 7 days (IgG3) to 21 days (IgG1, IgG2 and IgG4). Based on study on the Fc amino acid residue substitution it is suggested that the binding site of the antibody Fc and FcRn is present in the interface of CH2 domain and CH3 domain (Non-patent Document 27).
Based on the above, the human IgG1 antibody is most suitable as a therapeutic antibody since it has higher ADCC activity and CDC activity and further has longer half-life in human blood than other subclasses.
C1q is known to bind to the Fc region of the antibody molecule. The binding constants (Ka) in the binding of C1q to a monomeric antibody molecule in human IgG1, IgG2, IgG3 and IgG4 are 1.2×104, 0.64×104, 2.9×104 and 0.44×104 M−1, respectively (Non-patent Document 28). As discussed the above, especially CH2 domain among the Fc region is important (Non-patent Document 29).
Further in detail, regarding EU index (Non-patent Document 30), it is known that Leu 235 (Non-patent Document 31), Asp 270, Lys 322, Pro 329 and Pro 331 (Non-patent Document 32) in the CH2 are important in human IgG1, and Glu 233, Leu 234, Leu 235, Gly 236 (Non-patent Document 33), and Lys 322 (Non-patent Document 34) are important for human IgG3.
Attempts to enhance CDC activity by preparing a IgG variant by introducing amino acid substitutions in human IgG heavy chain constant region, and thereby increasing its binding activity to C1q.
Idusogie et al. reported that CDC activity is enhanced about 2-fold at the maximum when Lys at position 326 or Glu at position 333 according to the EU index in the CH2 domain of the heavy chain constant region of an anti-CD20 chimeric antibody Rituxan (registered trademark) comprising a human IgG1 type constant region and a mouse-derived variable region is substituted with other amino acid (Non-patent Document 35, Patent Document 2).
Idusogie et al. further showed that the CDC activity of IgG2 which was about one per several hundreds of the CDC activity of IgG1 increases to about 1/25 of the CDC activity of IgG1 when Lys at position 326 or Glu at position 333 according to the EU index in the human IgG2 type antibody is substituted with other amino acid (Patent Documents 3 to 5).
In addition, Dall'Acqua et al. reported that by applying various amino acid substitutions to the hinge region of a human IgG1 type anti-EphA2 antibody, plural of variants are exhibit higher C1q binding activity and higher CDC activity than those of the antibody before the substitutions (Non-patent Document 35).
Different from the techniques for introducing amino acid substitutions, an example in which CDC activity is enhanced by a combination of naturally existing sequences is also known. Shitara et al. found that when the full CH2 domain of human IgG1 and the full CH3 domain or a part of its N terminal side of human IgG1 are substituted with the sequence of human IgG3, the C1q binding activity and the CDC activity are greatly increased than those of IgG1 and IgG3 (Patent Document 6, Non-patent Document 36).
Two types are existed in the sugar chains bound to a protein, that is, N-glycoside-linked sugar chain and O-glycoside-linked sugar chain. The sugar chain which is linked to the N atom of amide group in the side chain of Asn in the protein is N-glycoside-linked sugar chain, and the sugar chain which is linked to the 0 atom of a hydroxyl group in the side chain of Ser and Thr in the protein is O-glycoside-linked sugar chain.
The N-glycoside-linked sugar chain includes three types, that is, a high mannose type sugar chain, a complex type sugar chain and a hybrid type sugar chain.
The process of biosynthesis of N-glycoside-linked sugar chain starts in the rough-surfaced endoplasmic reticulum from the attachment of dolichol-pyrophosphoric acid-oligosaccharide containing 9 molecules of mannose (Man) [(Glc)3(Man)9(GlcNAc)2] to the Asn residue of a consensus sequence Asn-X-Ser/Thr (X is any amino acid other than proline) of the N-glycoside-linked sugar chain. Thereafter, the sugar chain intermediate is converted into high mannose types such as Man8 type, Man7 type, Man6 type and Man5 type by various enzymes to synthesize complex type sugar chains to which N-acetylglucosamine, galactose, sialic acid, fucose and the like, instead of mannose, is linked (Non-patent Document 37).
However, it is known that the N-glycoside-linked sugar chain is not linked to all of the Asn-X-Ser/Thr sequences in a protein and is not linked further depending on the peripheral amino acid sequences and three-dimensional structure. Particularly, it is known that the addition of sugar chain hardly occurs to the Asn residue of Asn-X-Ser/Thr when the amino acid next to the C-terminal side of Asn-X-Ser/Thr is Pro (Non-patent Documents 38 and 39).
It is known that a complex-type N-glycoside-linked sugar chain binds to only Asn at position 297 of the CH2 domain in the constant region of the human IgG (Non-patent Document 40), however it is not known that the sugar chain binds to other position. It is known only that extra N-glycoside-linked sugar chain is linked to Asn at position 471 of constant region of mouse IgG3 (Non-patent Document 41).
On the other hand, a sequence of Asn-Thr-Thr is present at positions 392 to 394 of the CH3 domain in the human IgG3 type constant region, but the amino acid at position 395 is Pro (Non-patent Document 42), and actually, addition of sugar chain to Asn at position 392 of IgG3 is not known.
It is known that ADCC activity of human IgG is changed depending on the structure of the complex-type N-glycoside-linked sugar chain to be added to the Asn at position 297 (a typical schematic illustration of complex-type sugar chain is shown in FIG. 1A and FIG. 1B) (Patent Document 7).
Although there are reports stating that the ADCC activity of antibodies changes depending on the containing amounts of galactose and N-acetylglucosamine in the sugar chains to be linked to the antibodies (Non-patent Documents 43 to 46), a substance which exerts most influence upon ADCC activity is the fucose that performs α1,6 bond to the reducing terminal of N-acetylglucosamine (hereinafter referred to as a core fucose in some cases).
An IgG antibody comprising a complex-type N-glycoside-linked sugar chain having no core fucose shows markedly higher ADCC activity than that of an IgG antibody comprising complex-type N-glycoside-linked sugar chain having core fucose (Non-patent Documents 47 and 48, Patent Document 7).
As the cells which produce an antibody composition comprising complex-type N-glycoside-linked sugar chains having no core fucose, cells in which α1,6-fucosyltransferase gene was knocked out are known (Patent Documents 8 and 9).
All of FcγR-dependent activities such as ADCC activity, phagocytosis, CDC activity and FcRn binding activity are important for the therapeutic effect of a therapeutic antibody.
However, since all of the C1q binding which is the initial stage of inducing CDC activity, binding to FcγR which is the initial stage of inducing ADCC activity and binding to FcRn which contributes to the long blood half life are effected via the antibody Fc region of the antibody, there is a possibility that these activities are decreased when an amino acid substitution is introduced into the Fc region of the antibody.
Idusogie et al. reported that ADCC activity is greatly decreased in the case of introducing an amino acid point mutation into the CH2 domain of CDC activity-enhanced IgG1 (Non-patent Document 49).
Also, Dall'Acqua et al. reports a phenomenon that introduction of amino acid mutation into the CH2 domain of IgG1 for increasing binding activity to FcRn decreases ADCC activity at the same time (Non-patent Document 50).
Interestingly, these reports show that although the binding sites to FcγRIIIa, C1q and FcRn in the Fc are positioned at slightly different site, the amino acid substitution for changing each binding activity has unexpected influences upon other binding sites.
The amino acid at position 392 of the IgG heavy chain constant region is positioned at the interface of two CH3 domains in the antibody molecule (Non-patent Document 51). Since the amino acid positioned at the interface plays an important role in the association of two heavy chain molecules (Non-patent Document 52), for example, when the amino acid at position 392 is modified, there is a possibility that it has unexpected influences upon the three-dimensional structure and various biological activities of the antibody molecule.